Environ. Sci. Technol. 2009, 43, 7310–7316
Seasonal and Daily Source Apportionment of Polycyclic Aromatic Hydrocarbon Concentrations in PM10 in a Semirural European Area BAREND L. VAN DROOGE AND ´ REZ BALLESTA* PASCUAL PE European Commission - Joint Research Centre, Institute for Environment and Sustainability, TP-441, Ispra (VA) 21027 Italy
Received May 27, 2009. Revised manuscript received August 5, 2009. Accepted August 13, 2009.
Polycyclic aromatic hydrocarbons (PAH) were analyzed from ambient air particulate matter (PM10) that was collected near Lago Maggiore (Lombardy) in Northern Italy from August 2008 to January 2009. Highest individual PAH concentrations ranged from 0.05 ng/m3 during mid-days in summer to about 6 ng/m3 during the nights of the coldest period. A multivariate experimental regression model for the estimation of PAH concentrations was used to apportion the identified local sources of PAHs. This model included specific markers (i.e., levoglucosan for wood combustion and hopanes for traffic emission) and meteorological parameters (i.e., ambient air temperature and atmospheric mixing layer heights). In autumn and winter, wood combustion on a daily average contributed from 30 to 70% to the PAHs in ambient air PM10. In this period, the contribution of the wood combustion was less than 30% during mid-day, increasing sometimes to more than 80% during the night. In the samples taken during the summer period, traffic contribution to PAH concentrations was about 30%, while wood combustion was insignificant.
1. Introduction There is a causal relationship between particulate matter (PM) in the atmosphere and adverse health effects in humans (1, 2). Polycyclic aromatic hydrocarbons (PAHs) make up an important part of the toxic fraction of the particulate phase in ambient air (3). They are permanently formed by all sorts of incomplete combustion processes, such as fuel combustion in vehicles (4-6) or wood combustion (7-10). Therefore, PAHs may be considered ubiquitous pollutants, and human and environmental exposure to them is unavoidable. Air quality standards have been implemented to safeguard human health against the adverse effects of PAHs, and benzo[a]pyrene, in specific (11). However, a first step in the air quality evaluation, and eventually improvement of its quality, is the monitoring of the ambient air and the identification of potential emission sources. Although only 16% of the air quality monitoring stations in the European Union are located in rural or remote areas (12), it is important to better understand the air quality in semirural areas (between 100 and 500 inhab/km2) since approximately 135 * Corresponding author telephone: +390332785322; +390332789364; e-mail:
[email protected]. 7310
9
fax:
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 19, 2009
million people live in this type of area within the European Union (29% of the total population) (13). Moreover, contaminants emitted in urban and industrial zones can affect these areas after middle- and long-range atmospheric transport. Furthermore, they are frequently characterized by wood combustion for domestic heating and pollution episodes from transient combustion of biomass, i.e., open fires in gardens and fields. This is the case for the studied site in a semirural prealpine area, close to Lago Maggiore in the Lombardy region in Northern Italy (S1 in the Supporting Information). A recent survey (14) estimated that 30% of the PM10 emissions in the Lombardy region was attributed to wood combustion and 35% to traffic. As the emission rates of PAHs from wood combustion (3.7-13.7 mg/kg) (10) are higher than those from fossil fuel combustion in vehicles (0.008-0.240 mg/kg) (4, 5) the apportionment of wood combustion in semirural areas assumes importance. Moreover, the area is susceptible to strong temperature inversions during the year, especially in winter, which could worsen the air quality. A common method for identification of emission sources for PAHs in PM is the evaluation of diagnostic ratios of isomeric pairs and PAH profiles using conventional receptor models, such as chemical mass balances (CMB), principal component analysis (PCA), UNMIX, or positive matrix factorization analysis (15-20). These methodologies assume a unique profile of PAHs for each specific source and conservation of this PAH profile between the emission source and the measurement site. These two assumptions have been questioned, since there is a large variability in reported PAH profiles or diagnostic ratios and the influence of processes related to the physiochemical properties of PAHs, such as volatility, water solubility and photochemical reactivity with oxidants, which alter the atmospheric composition after emission ((21) and references therein). In addition, back trajectories of air mass to determine the source regions and the influence of middle- and long-range atmospheric transport on atmospheric concentrations has also been used in areas where few potential point sources were well-known (22). Nevertheless, this technique has its limitations in zones where the emission sources are not punctually localized (major sources) and, instead, minor sources are widespread and diffused through the area, as is the case in our study ((23) and references therein). On considering the limitations of conventional receptor models and taking into account the characteristics of the area in terms of potential emission sources, location, and meteorology, a simplified model based on a multivariate correlation of a limited number of PAHs, representative emission markers, and meteorological and atmospheric dispersion indicators was applied for source apportionment. In the present study, ambient air PM10 samples were collected from summer to winter for PAH quantification. Besides the conventional sampling strategies of 24 h, samples of 3 h were collected to take advantage of the potential variability of source emissions and meteorological conditions during a day. Furthermore, markers for the potential sources, such as levoglucosan in the case of wood combustion, and hopanes in the case of (diesel) vehicle emissions (6, 7) were identified and quantified in the samples for source apportionment purposes.
2. Experimental Section The sampling, analytical methodology, and meteorological conditions during the campaign are described in more detail elsewhere ((24) and Supporting Information (S2.1 and S2.2)). 10.1021/es901381a CCC: $40.75
2009 American Chemical Society
Published on Web 08/27/2009
Briefly, low volume samplers (2.3 m3/h) were used for sampling PM10 during 24 and 3 h on quartz filter following the standardized method described in EN 12341 (25). The sampling covered the seasonal transition from August 2008 to January 2009 in a semirural background area of Northern Italy at the Joint Research Centre (JRC) in Ispra (S1). Samples were analyzed directly by thermal desorption and gas chromatography mass spectrometry. Sixteen PAHs, ranging from phenanthrene to benzo[ghi]perylene (24), together with levoglucosan, retene, and hopanes were quantified in the PM10 fraction for source apportionment purposes. In total 28 samples of 24 h and 68 samples of 3 h sampling were analyzed. Meteorological data were obtained from the EMEP station located at the JRC Ispra site and air mass back-trajectories and estimations of the mixing layer heights during the samplings were calculated by means of the HYSPLIT4 software (26). Average daily temperature ranged from 19 to 21 °C in the warm period (August-September), moving to average temperatures ranging from 9 to 15 °C between September and November (intermediate period), to reach lower values ranging from -2 to 6 °C after middle November. In addition, frequent episodes of prolonged temperature inversion characterized the cold period from November to the end of January. As a consequence of lowering ambient temperatures and less solar radiation toward the winter season, the studied period shows a decline of the average mixing height, with an average height above ground level ranging from 1136 to 1668 m in the warm period, from 169 to 449 m in the intermediate period, and from 80 to 278 m in the cold period (S2.3, S2.4).
3. Results and Discussion The complete data set of the measurements is presented in S3 and S4 of the Supporting Information. Measured concentrations of the individual PAHs were all above the limit of detection. Nevertheless, anthracene and perylene were below their limit of quantification in samples collected in August and September (2-3/9 and 7-8/9). 3.1. Seasonal and Daily Variation on the PAH Concentration Levels. All PAHs showed lowest concentrations in the warm period, with maximum concentrations of individual PAHs of about 0.05 ng/m3. The mean daily concentration in the intermediate period from September to November ranged from 0.03 ng/m3 for anthracene to 0.6 ng/m3 for benzo[b+j]fluoranthene, while the mean concentrations in the cold period ranged from 0.09 for anthracene to 2.65 ng/m3 for benzo[b+j]fluoranthene (S3). PAH concentrations in the warm period were comparable to those detected in remote areas (27, 28), while those detected in the cold period were comparable to those observed in urban background sites (22, 29). In particular, average daily concentrations of the regulated benzo[a]pyrene were around 0.01 ng/m3 in summer, while they increased to maximum levels of 4 ng/m3 in the cold period. The mean concentration of benzo[a]pyrene over the analyzed period was 0.8 ng/m3, which is close to the annual limit value of 1 ng/m3 established in the European legislation (11). Although the analyzed period does not represent a complete mandatory sampling period, it is a clear indication of a poor air quality in the area, especially during winter. The presence of higher PAH concentrations (S3) in colder periods in comparison to those measured in warmer periods was in agreement with other studies and was explained by the combination of factors related to the physicochemical properties, such as volatility related to the particle gas partitioning, and reactivity with oxidants, but also with meteorological conditions, such as atmospheric stability, and the strength of the potential emission sources (22, 29-31).
The PAHs show a wide range of volatilities (32), which condition the effect of temperature on the particle-gas equilibrium. As in the present study only particulate PAHs were analyzed, the effect of temperature variations on the particle-gas partitioning distribution can only be theoretically estimated on the basis of previous research. Therefore, according to experimental and theoretical data of the particle-gas partitioning coefficient (Kp) ((33) and references therein), the particulate phase concentrations of phenanthrene and benzo[a]pyrene were estimated under our experimental conditions of being 2.8 and 2.1 times higher in the cold period with respect to the warm period, respectively. Moreover, based on the temperature variations during a day, the daily estimated particle concentrations for phenanthrene and benzo[a]pyrene would vary around a factor 1.5. Nevertheless, in the present study the observed particle phase concentrations for phenanthrene and benzo[a]pyrene were, respectively, 20 and 140 times higher in the cold period with respect to the warm period. Whereas the observed variations during the day were between a factor 1.5 and 2.6 for phenanthrene and between 5.6 and 13.7 for benzo[a]pyrene. In spite of the negative relationship (R2 ) 0.78, p < 0.001) between ambient temperatures and PAH concentrations in PM10 (S5), the temperature dependence of the particle-gas partitioning of the PAHs cannot fully explain the observed seasonal and daily PAH concentration changes in the ambient PM10. Meteorological parameters related to atmospheric dispersion may also affect concentration levels of PAHs. As mentioned in the introduction, the area studied is susceptible to prolonged periods of high atmospheric stability, with nocturnal stable layers (NSL) formed at ground level due to calm winds and the particular orography of the Po valley. This phenomenon, that is affecting the whole region, was studied by other researchers (34), who related variation of radon concentrations with temperature inversions, demonstrating that temperature inversion is a good atmospheric dispersion predictor for this area. As an indicator of temperature inversion, mixing layer heights were calculated for all sampling days (S2.4). For most of these days, the highest mixing heights were around midday and the lowest at night and, in general, a steep decrease of the mixing heights occurred after sunset, when atmospheric temperatures drop to increase again after sun rise. A negative correlation (R2 ) 0.64, p < 0.001) between mixing heights and PAH concentrations was observed (S6). Nevertheless, the daily and seasonal PAH variations were much larger than those expected from the typical atmospheric dispersion in the area, i.e., factor 4, derived from variations on radon concentration (34). During atmospheric transport, PAH compounds are exposed to O3, OH, and NO2 and other potential PAH photooxidants that react and change PAH relative composition (3, 35-39). Compounds like anthracene (ANT), pyrene (PYR), benzo[a]pyrene (BaP), benzo[a]anthracene (BaA), and benzo[ghi]perylene (BGP) seem to be more susceptible to oxidation or photodegradation than their corresponding isomers: phenanthrene (PHE), fluoranthene (FL), benzo[e]pyrene (BeP), chrysene+triphenylene (CHR), and indeno[123 cd]pyrene (IP) (15, 39, 40). For this reason, ratios of these isomers cannot be used as a stand alone source apportionment tool (15-17). In our case, ratios of PHE/(PHE+ANT) and FL/(FL+PYR) were slightly higher during August-October period when compared to those from November-January, the opposite was, however, markedly observed for the ratios BaA/ (BaA+CHR), BaP/(BaP+BeP) and IP/(IP+BGP), which agreed with the expected reactivity of these compounds. In general, the scattering of these ratios during the warm period was higher than that from the winter period (see Table 1 and S7, S8). Nevertheless, this is also in concomitance with seasons VOL. 43, NO. 19, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
7311
TABLE 1. Characteristic PAH Isomer Concentration Ratios during the Warm and Cold Periodsa isomer concentration ratio
warm period
cold period
t test for independent samples
mean (STDEV)
mean (STDEV)
t
p
PHE/(PHE+ANT) FL/(FL+PYR) BaA/(BaA+CHR) BaP/(BaP+BeP) IP/(IP + BGP)
0.86 (0.03) 0.52 (0.03) 0.21 (0.06) 0.30 (0.07) 0.31 (0.06)
0.82 (0.04) 0.47 (0.02) 0.35 (0.03) 0.51 (0.06) 0.50 (0.03)
4.43 5.00 -9.49 -11.00 -11.76
0.00018 0.00013
